2022

• Record 73 of
  
  Title:Alzheimer's level classification by 3D PMNet using PET/MRI multi-modal images
  
  Author(s):Li, Chao(1,2,3); Song, Liyao(4); Zhu, Guangpu(1,2,3); Hu, Bingliang(1,3); Liu, Xuebin(1,3); Wang, Quan(1,3)
  Source: 2022 IEEE International Conference on Electrical Engineering, Big Data and Algorithms, EEBDA 2022  Volume:   Issue:   DOI: 10.1109/EEBDA53927.2022.9744769  Published: 2022  
  Abstract:The accurate diagnosis of Alzheimer's disease (AD) has an important impact on early treatment. Positron emission tomography (PET) and magnetic resonance imaging (MRI) are popular imaging methods and are used to facilitate the identification and evaluation of AD. In this paper, we proposed a VGG-style 3D convolutional neural network (3D CNN) model, which is named 3D PET-MRI Net (3D PMNet), and it uses DiffGrad optimizer to speed up the convergence of the model and Focalloss function to improve the classification performance of unbalanced data processing. The multi-modal feature information of 3D MRI and PET images can be extracted using the 3D PMNet model, which provides convenience for AD diagnosis. Tenfold cross-validation was performed on the data of each patient in the data set to determine the group classification. The results showed that the proposed method achieves 97.49%, 81.25%, and 76.67% accuracy in the classification tasks of AD: NC, AD: MCI, and NC: MCI, respectively. Our PMNet reached 72.55% accuracy in AD: NC: MCI three group classification, which is significantly better than the other reported network models. © 2022 IEEE.
  Accession Number: 20221712027361
• Record 74 of
  
  Title:Two-Directional Two-Dimensional PCA: An Efficient Face Recognition Method for Thermal Infrared Images
  
  Author(s):Gao, Chi(1,2); Zhang, Xinming(1,2); Wang, Hui(1,2); Song, Liyao(3); Hu, Bingliang(1); Wang, Quan(1)
  Source: 2022 5th International Conference on Information Communication and Signal Processing, ICICSP 2022  Volume:   Issue:   DOI: 10.1109/ICICSP55539.2022.10050541  Published: 2022  
  Abstract:Compared with face recognition in the environment of visible light, thermal infrared face recognition has the advantages of being independent of light, working around the clock, and capable of detecting hidden targets easily. In this paper, we propose a thermal infrared face recognition method based on the two-directional two-dimensional PCA (2D2DPCA) and random forest classifier. We compared this with two deep learning networks: Alexnet, Three-dimensional Convolutional Neural Networks (3DCNN), and applied these with two databases: the Terravic Facial IR database (with different facial angles) and the NVIE database (with various emotional expressions). Among these methods, the accuracy of face recognition with the 2D2DPCA method achieves the best recognition effect, it reached 99.92% and 99.97% in both databases, respectively. We statistically verified that our method could not only accurately and robustly recognize thermal infrared faces with large variations in angle and expression, but also greatly reduce computational complexity and data dimension, improving the speed of face recognition. With the two sample sets tested, our work has demonstrated that 2D2DPCA has excellent potential for facial image compression and may broaden thermal face recognition applications. © 2022 IEEE.
  Accession Number: 20231113742344
• Record 75 of
  
  Title:Image Enhancement Technology in Pavement Disease Detection System
  
  Author(s):Li, Xuefeng(1); Zhou, Zuofeng(2); Wu, Qingquan(2)
  Source: 2022 IEEE 2nd International Conference on Electronic Technology, Communication and Information, ICETCI 2022  Volume:   Issue:   DOI: 10.1109/ICETCI55101.2022.9832258  Published: 2022  
  Abstract:Efficient pavement bad location detection and repair is essential to prolong the use time of roads. However, traditional manual detection methods are extremely inefficient and can no longer meet the requirements of inspecting a large number of roads. When using deep learning technology for road disease detection, it is found that low-illuminance images will affect the detection accuracy due to low contrast. Therefore, before training and testing the deep learning model, the original image needs to be preprocessed to improve the image quality. First, bilateral filtering is used instead of Gaussian filtering to estimate the illuminance of the original image; Then the reflection component is get according to the principle of Retinex algorithm, and the reflection image is quantized; Finally, the image is subjected to illumination compensation. The results of comparative experiments display that the ours algorithm can retain the characteristic details of road diseases and eliminate the unevenness of the image brightness distribution while improving the contrast of the road image. © 2022 IEEE.
  Accession Number: 20223312571189
• Record 76 of
  
  Title:Spectral Beam Combing of Fiber Lasers with 32 Channels
  
  Author(s):Gao, Qi(1,2); Li, Zhe(1,2); Zhao, Wei(1); Li, Gang(1,2); Ju, Pei(1,2); Gao, Wei(1,2); Dang, Wenjia(3)
  Source: SSRN  Volume:   Issue:   DOI: 10.2139/ssrn.4291145  Published: December 1, 2022  
  Abstract:We present a method for spectral combination of fiber lasers with extremely high spectral density, increasing spectral density utilization with no degradation in beam quality, and decreasing the single channel narrow linewidth output power. Experiments demonstrating the utility of our method are described. The results show that we achieve 32 channels fiber laser spectral beam combining (SBC) with a beam quality of M2 =1.68. The beam quality of SBC can be optimized constantly by varying the spectral interval integrally with the feedback system. Our method is potentially scalable to many 100’s of channels and achieves tens or hundreds of kW output power with an excellent beam quality. © 2022, The Authors. All rights reserved.
  Accession Number: 20220449368
• Record 77 of
  
  Title:10-W Random Fiber Laser Based on Er/Yb Co-Doped Fiber
  
  Author(s):Li, Zhe(1,2); Gao, Qi(1,2); Li, Gang(1,2); She, Shengfei(1,2); Sun, Chuandong(1); Ju, Pei(1,2); Gao, Wei(1,2); Dang, Wenjia(3)
  Source: SSRN  Volume:   Issue:   DOI: 10.2139/ssrn.4291140  Published: December 1, 2022  
  Abstract:In this study, we presented a 1550 nm, high-power, high-efficiency random fiber laser. A method, utilizing the single-mode erbium-ytterbium co-doped fiber with proper length and the highly reflective fiber Bragg grating with wide reflection bandwidth, is used to surmount the generation of Yb-ASE and low slope efficiency. More than 10 W output power is achieved, with a slope effi-ciency of 36.7% and single transverse mode output. The random fiber laser stably operates without significant amplitude fluctuation under maximum power, and which can provide a high-performance light source for a variety of applications. © 2022, The Authors. All rights reserved.
  Accession Number: 20220449283
• Record 78 of
  
  Title:Chinese Character Font Classification in Calligraphy and Painting Works Based on Decision Fusion
  
  Author(s):Zeng, Zimu(1,2); Zhang, Pengchang(1); Wang, Jia(3); Tang, Xingjia(1); Liu, Xuebin(1)
  Source: Proceedings - 2022 IEEE/WIC/ACM International Joint Conference on Web Intelligence and Intelligent Agent Technology, WI-IAT 2022  Volume:   Issue:   DOI: 10.1109/WI-IAT55865.2022.00117  Published: 2022  
  Abstract:Font recognition is an important part in the field of painting and calligraphy style recognition. Traditional font classification methods are mainly based on texture feature extraction and other methods, which need to be improved in classification accuracy. The mainstream classification methods mainly use convolutional neural networks, but such methods have poor interpretability and may face the problem that some detailed features cannot be accurately extracted. Based on convolutional neural network, the gray-level images, Local Binary Pattern (LBP) feature and Histogram of Oriented Gradient (HOG) of the images in the font dataset are respectively trained. Finally, the results of the three networks are fused by means of average decision fusion. The experimental results of font recognition show that the proposed method can extract the detailed features of fonts more accurately and obtain higher classification accuracy. © 2022 IEEE.
  Accession Number: 20231914078169
• Record 79 of
  
  Title:Electronic image stabilization algorithm for space exploration based on star point extraction
  
  Author(s):Yanliang, Li(1,2); Yan, Wen(1); Dong, Wang(1); Wencan, Li(1)
  Source: Proceedings of SPIE - The International Society for Optical Engineering  Volume: 12169  Issue:   DOI: 10.1117/12.2624047  Published: 2022  
  Abstract:In deep space exploration, the optical system is susceptible to various factors in space, resulting in instability of the visual axis. In order to improve the imaging quality, high-precision optical axis pointing is required. This paper is designed to feed back the current optical axis pointing in real time during space exploration. Deviation algorithm. We use an improved threshold segmentation algorithm and secondary judgment to improve the accuracy of star point extraction, which can effectively extract star point pixels in real star images. Through the extracted star point pixels, we use a threshold-based gray square weighted centroid calculation method to calculate the centroid of the star point, and use the centroid deviation of the navigation star point to obtain the final optical axis pointing deviation. In addition, we also use the windowing method to speed up the calculation rate after obtaining the navigation star point. Experiments show that the algorithm can feedback the optical axis deviation of the optical system in real time. © 2022 SPIE
  Accession Number: 20221611967882
• Record 80 of
  
  Title:Study on the Influence of Deposition Temperature on the Properties of Lanthanum Titanate Films
  
  Author(s):Li, Yang(1); Xu, Junqi(1); Su, Junhong(1); Liu, Zheng(2)
  Source: OGC 2022 - 7th Optoelectronics Global Conference  Volume:   Issue:   DOI: 10.1109/OGC55558.2022.10050984  Published: 2022  
  Abstract:The work aims to study the effect of deposition temperature on optical properties and residual stresses in Lanthanum titanate (H4) films. The LaTiO3 films were deposited by electron-beam thermal evaporation technique. The residual stress of LaTiO3 films on fused silica was characterized macroscopically and microscopically, using laser interferometry and AFM. The residual stresses and surface profile shape change were simulated using finite element analysis methods. It was confirmed that the deposition temperature did not affect the optical properties of the films but did for residual stresses. The residual stress of LaTiO3 films changes from decreasing tensile stress to compressive stress as the deposition temperature increases. The deposition temperature is used to modulate the magnitude and transition of the residual stress in the films. There is a strong dependence between the residual stresses and the densities of surface columnar structures in LaTiO3 films. The effect of density of surface columnar structures is found as follows: the film with the lower density of surface columnar structures generally shows a tensile and high density easily transform into compress stress. This conclusion is also verified by the increase of the corresponding refractive index. The simulated surface profiles are basically overlapping with the measured data. The proposed model is validated for the simulation of residual stresses in monolayers. © 2022 IEEE.
  Accession Number: 20231113708384
• Record 81 of
  
  Title:ReIMOT: Rethinking and Improving Multi-object Tracking Based on JDE Approach
  
  Author(s):Hou, Haoxiong(1,2); Zhang, Ximing(3); Sun, Zhonghan(3); Gao, Wei(3)
  Source: 2022 5th International Conference on Pattern Recognition and Artificial Intelligence, PRAI 2022  Volume:   Issue:   DOI: 10.1109/PRAI55851.2022.9904121  Published: 2022  
  Abstract:The multi-object tracking (MOT) algorithms of the joint detection and embedding (JDE) approach estimate bounding boxes and re-identification (re-ID) features of objects with the single network, which balance the tracking accuracy and inference speed. However, when the appearance information between different objects is highly similar, these algorithms are usually easy to cause identity switches, and the comprehensive tracking performance is poor in crowded scenes. Aiming at the above problems, we propose a stronger multi-object tracking algorithm termed as ReIMOT, based on FairMOT. A joint loss function of combining normalized Softmax Loss and the center distance penalty term is designed to supervise the re-ID branch, which increases the intra-class similarity and makes the extracted appearance features more discriminative. To further improve the tracking performance, we introduce coordinate attention to make the encoder-decoder network focus more on features of interest. The experimental results show that the proposed ReIMOT is more effective than the other advanced multi-object tracking algorithms, and decreases the number of ID switches by 13.8% compared to FairMOT on the MOT17 dataset. © 2022 IEEE.
  Accession Number: 20224513060941
• Record 82 of
  
  Title:Analysis and experiment of small target detection in high speed flow field of near space
  
  Author(s):Guo, Huinan(1); Ma, Yingjun(1); Wang, Hua(1); Peng, Jianwei(1)
  Source: Hongwai yu Jiguang Gongcheng/Infrared and Laser Engineering  Volume: 51  Issue: 12  DOI: 10.3788/IRLA20220218  Published: December 2022  
  Abstract:With the deepening of space security and application exploration, the target-detectability of space vehicle in near space has become a core issue of research. For some multi-dimensional information of target, such as shape, spectrum and motion characteristics, can be directly captured by optical imaging detection device, optical detection has become an important means of space imaging and target detection. Under the conditions of atmospheric density, pressure and atmospheric convection in near space, imaging quality and detection range of optical detection device installed in high-speed aircraft could be affected seriously. By using target detection model with three analysis elements (imaging system, atmospheric transmission system and target-background system) and the theory of aero-optical effect, evaluation equation of aero-optical effect for high speed flow field has been established, to analyze imaging performance of typical scenes such as earth and space background. A ground verification test of target detection in high speed flow field has also been designed. The experimental results show that it’s an effective way for detecting plume flow of high-speed space targets by using short wave infrared detector (SWIR: 900-1 700 nm) with quartz window (with thickness of more than 10 mm). Meanwhile, by reducing exposure time of camera, optimizing exposure control strategy and selecting optical filter, stray light in background and aero-optical effect can be effectively suppressed. © 2022 Chinese Society of Astronautics. All rights reserved.
  Accession Number: 20230213368779
• Record 83 of
  
  Title:Influence of the Rotary Ultrasonic Vibrating Direction on Surface Quality in Aspheric Grinding Glass-Ceramics
  
  Author(s):Sun, Guoyan(1,2); Shi, Feng(1); Zhang, Bowen(3); Zhao, Qingliang(3); Zhang, Wanli(1); Wang, Yongjie(2); Tian, Ye(1)
  Source: SSRN  Volume:   Issue:   DOI: 10.2139/ssrn.4119791  Published: May 26, 2022  
  Abstract:Glass-ceramics are considered superior materials for aspherical optics in large-aperture telescopes and space mirrors due to their outstanding mechanical and thermal performance. To improve the processing quality and efficiency of glass-ceramics, ultrasonic vibration assisted grinding (UVG) is widely studied, focusing on machining mechanism and surface generation. However, the machining characteristics of aspheric surface are rarely studied. Herein, rotary ultrasonic vibration assisted vertical grinding (RUVG), where the vibration direction of grinding wheel is parallel to the rotation liner velocity direction of the workpiece, and rotary ultrasonic vibration assisted parallel grinding (RUPG), where the vibration direction of grinding wheel is vertical to the rotation liner velocity direction of workpiece, are proposed for aspheric surface machining of glass-ceramics. To reveal the surface formation mechanism of both UVG methods theoretically, single-grain kinematic functions are created and contact characteristics between the grinding wheel and aspheric surface are analyzed, as well as the grinding marks corresponding to RUVG and RUPG are simulated. It is worth noting that different ultrasonic vibration (UV) directions lead to significant differences in cutting contact time, contact area, instantaneous relative velocity value and velocity direction between the aspheric surface and grinding wheel. Subsequently, comparative experiments are conducted on an ellipsoid surface of glass-ceramics and the results indicate that there are slight distinctions in macro-grinding surface texture pattern and surface roughness between RUVG and RUPG. From the surface form accuracy viewpoint, RUVG exhibits a more prominent influence than the RUPG, rendering a low surface profile error. The differences in grinding surface quality of RUVG and RUPG mainly depend on grinding parameters, UV parameters and material properties. The current research enables an in-depth understanding of comprehensive mechanisms of RUG for aspheric surface machining of brittle materials and provides theoretical bases for the application of UVG methods on the machining of complex surfaces. © 2022, The Authors. All rights reserved.
  Accession Number: 20220121467
• Record 84 of
  
  Title:NTIRE 2022 Spectral Recovery Challenge and Data Set
  
  Author(s):Arad, Boaz(1,2); Timofte, Radu(3); Yahel, Rony(1,4,5); Morag, Nimrod(1,2,6); Bernat, Amir(1,2); Cai, Yuanhao(7); Lin, Jing(7); Lin, Zudi(8); Wang, Haoqian(7); Zhang, Yulun(9); Pfister, Hanspeter(7); Van Gool, Luc(8); Liu, Shuai(10); Li, Yongqiang(10); Feng, Chaoyu(10); Lei, Lei(10); Li, Jiaojiao(11); Du, Songcheng(11); Wu, Chaoxiong(11); Leng, Yihong(11); Song, Rui(11); Zhang, Mingwei(12); Song, Chongxing(13); Zhao, Shuyi(13); Lang, Zhiqiang(13); Wei, Wei(13); Zhang, Lei(13); Dian, Renwei(14); Shan, Tianci(14); Guo, Anjing(14); Feng, Chengguo(14); Liu, Jinyang(14); Agarla, Mirko(14); Bianco, Simone(15); Buzzelli, Marco(15); Celona, Luigi(15); Schettini, Raimondo(15); He, Jiang(16); Xiao, Yi(16); Xiao, Jiajun(16); Yuan, Qiangqiang(16); Li, Jie(16); Zhang, Liangpei(17); Kwon, Taesung(18); Ryu, Dohoon(18); Bae, Hyokyoung(18); Yang, Hao-Hsiang(19); Chang, Hua-En(19); Huang, Zhi-Kai(19); Chen, Wei-Ting(22); Kuo, Sy-Yen(21); Chen, Junyu(20); Li, Haiwei(20); Liu, Song(20); Sabarinathan, Sabarinathan(23); Uma, K.(24); Bama, B Sathya(24); Roomi, S. Mohamed Mansoor(24)
  Source: IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops  Volume: 2022-June  Issue:   DOI: 10.1109/CVPRW56347.2022.00102  Published: 2022  
  Abstract:This paper reviews the third biennial challenge on spectral reconstruction from RGB images, i.e., the recovery of whole-scene hyperspectral (HS) information from a 3-channel RGB image. This challenge presents the "ARAD_1K"data set: a new, larger-than-ever natural hyperspectral image data set containing 1,000 images. Challenge participants were required to recover hyper-spectral information from synthetically generated JPEG-compressed RGB images simulating capture by a known calibrated camera, operating under partially known parameters, in a setting which includes acquisition noise. The challenge was attended by 241 teams, with 60 teams com-peting in the final testing phase, 12 of which provided de-tailed descriptions of their methodology which are included in this report. The performance of these submissions is re-viewed and provided here as a gauge for the current state-of-the-art in spectral reconstruction from natural RGB images. © 2022 IEEE.
  Accession Number: 20223712740884